[petsc-dev] [petsc-maint #43201] Questions about TS functionality.

[Michael E Henderson]

Barry, Jed,

Thanks. That's good to hear. I'll give it a look.

Mike
------------------------------------------------------------------------------------------------------------------------------------
Mathematical Sciences, TJ Watson Research Center
mhender at watson.ibm.com
http://www.research.ibm.com/people/h/henderson/
http://multifario.sourceforge.net/




From:
Jed Brown <jed at 59A2.org>
To:
Barry Smith <bsmith at mcs.anl.gov>, petsc-maint at mcs.anl.gov, Michael E 
Henderson/Watson/IBM at IBMUS
Cc:
For users of the development version of PETSc <petsc-dev at mcs.anl.gov>
Date:
03/15/10 03:31 PM
Subject:
Re: [petsc-dev] [petsc-maint #43201] Questions about TS functionality.
Sent by:
Jed Brown <five9a2 at gmail.com>



Hi Michael, to elaborate on Barry's comments:

PETSc-dev (and the forthcoming 3.1 release) has support for DAEs using
the interface described here

  
http://www.mcs.anl.gov/petsc/petsc-as/snapshots/petsc-dev/docs/manualpages/TS/TSSetIFunction.html


Three of the TS implementations currently support this form:

TSGL: very new general linear methods (A- and L-stable, order 1-5 with 
full stage order, adaptive step and order)

TSTHETA: classic Theta method, Theta=0.5 is Trapezoid rule 
(Crank-Nicolson), Theta=1 is Backward Euler.

TSPSEUDO: pseudo-transient continuation


I wrote a section in the User's manual on solving DAE (6.1.2):

  
http://www.mcs.anl.gov/petsc/petsc-as/snapshots/petsc-dev/docs/manual.pdf


The controllers for TSGL are pretty naive so it will probably take a
while before they are especially robust.

As Barry says, we plan to make a number of improvements to TS (mostly in
implementation and extensibility) after the release.

Jed


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20100315/16d2eeb2/attachment.html>